Do you use complex methods to back / forward test?

Discussion in 'Strategy Development' started by amigasearch, Oct 21, 2003.

  1. Hi. When backtesting, or forward testing a strategy, all I do currently to determine its worth is, look at expectancy (look for a positive one) and make sure my wins overcome my losses (risk to reward). I also look for a nice equity curve, and minimal drawdown (less then 10 percent!). Finally, I will run my previous trades, scramble them up, and see results over 1000 iterations (randomly mixed). This seems to me to work.
    I notice strategy programs out there, provide some other, to me, complex ways, like effiency (I cant even spell it right!) and MAE.

    Are these parts neccesary? Do you use them now to make sure your strategy is worth trading, or do you cover all the bases (such as what your software provides).

    I am wondering before I begin trading a system that give me 50 / 50 win to loss ratio, but dollar amount on my wins outnumber my losses by 3:1 ratio. I am comfortable with this - would you be??
    And what can I use to verify somethings I am leaving out.
  2. The very first thing is to test your system on out-of-sample data. You can go back in time or forwad in time to do this.

    MAE can help you with stop management, among other things.(Explained very well in 'Campaign Trading')

    I would really reccomend abook like 'Design Testing and Optomization of Trading Systems' by Pardo, that can give you the skinny on these issues.

    Best Regards,